My dad used to keep a list to record each time he bought gas and the number of gallons purchased. When I finally asked why he was doing this, he explained that it helped him determine if his car was losing performance. By computing miles per gallon, he could determine if the car needed a tune-up or other adjustments to bring it back to peak performance. As a kid, I thought this was a waste of time. But as an adult, I learned to understand the need for accurate data and its implications.
Many companies that are interested in data collection have reasonably modern equipment, and pulling data from plant-floor processors has become relatively easy (but let’s keep that our little secret). Problems arise, however, when pulling data from old control systems.
System integrators can provide support with system modernization or special interfaces. But in order to get started, your integrator wants a basic understanding about the data to be collected, and sometimes these questions are deceptively difficult to answer:
- What data do you want to collect? The answer invariably is, “Everything.”
- What are you going to do with the data? The answer is almost always, “We’re not sure.”
- What questions do you want to answer with the data? See question 2.
If this is your first data collection project, my suggestion is to understand your infrastructure (to enable connecting all the “stuff”) and then develop the means to calculate overall equipment effectiveness (OEE). OEE is calculated as Availability x Performance x Quality. Each of these variables are easily calculated or derived from production data.
- Availability = Operating Time / Planned Production Time
- Performance = (Total Pieces / Operating Time) / Ideal Run Rate
- Quality = Good Pieces / Total Pieces
The initial result establishes the baseline. Then as OEE is measured over time, much like my dad did with his car, you can determine if the performance of your machine or process remains consistent, is improving, or is getting worse.
If you understand OEE and are using this value, then you’ve earned the right to dig deeper. From this same data, you can start looking at why the machine or process is not as available as it should be (downtime) or why your quality is slipping (good vs. bad). You will start to see what additional data you want to collect, what can be done with the data and what questions the data will answer.
Here’s the key to data collection: People must earn the data. Collect a small amount of data first and see if anyone uses it. If so, give them more. If not, don’t bother.
Some companies have a very sophisticated manufacturing execution system (MES) that can do things like compute and track OEE and many other metrics to determine plant performance as well as integrating production scheduling, recipe management and other tools such as diagnostics, reporting and analysis. That sounds pretty appealing, doesn’t it? Get started by collecting some simple data such as OEE. Doing so will provide the foundation for justifying a more complex MES. Getting started costs less than $10,000, so even a 1 percent improvement in OEE provides a quick return on investment.
Robert Lowe is co-founder and CEO of Loman Control Systems Inc., a certified member of the Control System Integrators Association (CSIA). See Loman’s profile on the Industrial Automation Exchange by CSIA.
